This project builds on three years of intensive participant-observation ethnographic fieldwork in the shooting encampments of a network of white and Latino homeless heroin addicts that documents the logics for HIV transmission due to syringe and ancillary paraphernalia sharing. The expansion of the core network to include a cohort of African American crack smokers and injectors has exacerbated the network's risk environment, and provides a comparative and evolving natural social laboratory that now permits the documentation in a fully indigenous setting of the ways risky injection practices articulate with: 1) ethnicity; 2) poly-drug abuse; 3) income-generating strategies; 4) gender/sexuality; and 5) status hierarchies. The restructuring of our network's boundaries and its dyadic running partnerships allows us to analyze the protective as well as risk-promoting dynamics of these socially structured relationships. We will identify network bridging and bounding mechanisms that shape the spread of HIV as well as offer contextualized processual understandings for why different individuals within the same network and across comparable networks engage in distinct patterns of risky injection behavior. A multi-ethnic research team will collect the data inside the shooting and sleeping encampments of the core network of 20 heroin injectors and their peripheral social contacts (approximately 50 additional injectors). The research design and analysis links theoretically significant categories addressing macro-social power relations to the specific observable risky behaviors of differentially vulnerable individuals. An analysis of the pragmatics and contexts for taboo behaviors that are sometimes concealed or misreported to outsiders offers public health service providers an indigenous perspective on the effectiveness of their outreach programs. The project engages a methodological dialogue without long-term epidemiological cohort studies of IDUs in order to operationalize qualitative data into testable epidemiological constructs. This ongoing iterative dialogue also allows us to reframe questions, generate new hypotheses, and systematize sampling strategies among out-of-treatment, hard-to-reach substance users. The long-term goal is to contribute to a socially contextualized model that explains variance in HIV infection/risk-behavior rates among different vulnerable sectors of the U.S. population by elucidating the social forces--both micro and macro -- that shape the spread of blood-bone disease among injectors.